Bayesian latent class models for prevalence estimation and diagnostic test evaluation in the absence of a gold-standard
November 17-18 2015
Faculty of Veterinary
Medicine, Université de Montréal
Nandini Dendukuri (UMcGill) and Ian Schiller (McGill Univ Health Centre) with the assistance of Simon Dufour (UdeM) and Sébastien Buczinski (UdeM)
Why this course?
During the last two decades Bayesian statistical methods have been widely applied in veterinary science and epidemiological research. In particular, Bayesian latent class models have been found to be useful for estimation of disease prevalence or diagnostic test accuracy (sensitivity and specificity) in the absence of a gold-standard reference test. This 2-day workshop will introduce these methods and illustrate their application to real-life examples from veterinary sciences. The format of the workshop will include both lecture and laboratory sessions giving participants the opportunity to learn how to apply these methods using WinBUGS software. The material will be suitable for graduate students or researchers in veterinary sciences, epidemiology, or public health with an interest in applications of novel statistical methods.
Who is this course for?
This course is for everyone with an interest in diagnostic test evaluation or prevalence estimation in the absence of a gold-standard test. While most of the examples used in the workshop will be animal health-derived examples, the methods presented are directly applicable to human health. Prior experience with animal health is not essential. Some basic epidemiological knowledge related to diagnostic test attributes (e.g. gold-standard, diagnostic sensitivity and specificity, prevalence of disease in a population) would be beneficial. The level of the course will be suitable for those who have had exposure to introductory, graduate-level courses in epidemiology and biostatistics.
The course will run from 8h30 to 17h00 on both days. Each half-day session will include a 1.5 hour lecture followed by a 1.5 hour computer lab separated by a 30 minute break. Lectures will be in English only, but support during the practical will be offered in French and English. Software that will be used for the computer lab is WinBUGS (freeware). Solutions for the exercises in SAS and R will also be available.